memax import

One-way push of supported files from a local directory into Memax. memax import does not choose a hub — there's no --hub flag, it doesn't read the active hub, and it doesn't read .memax.yml. With normal user credentials imports land in your personal hub; with hub-scoped API keys they land in whichever hub the key is bound to. Re-runs are idempotent (dedup by content hash).

Usage

memax import [directory]
memax import status

directory defaults to . (the current directory).

Options

Flag	Description	Default
`-w, --watch`	Watch for changes after initial import (experimental)	off
`-b, --boundary <level>`	Accepted for forward compatibility; recorded locally in the sync-source status only. Does not currently change server-side visibility (the server writes `boundary=private` regardless).	`private`
`--ignore <patterns>`	Additional comma-separated directory names to skip	—
`-y, --yes`	Skip the confirmation prompt when importing more than 10 files	off

Anything not listed above is not a recognized flag. There is no --include, --exclude, --hub, or --dry-run today.

Which files are imported

Only files whose extension matches the allowlist below are imported. Everything else (binaries, images, unknown extensions) is skipped silently without a warning.

.md  .txt  .ts  .tsx  .js  .jsx  .go  .py  .rs
.yaml  .yml  .json  .toml  .sh  .bash  .zsh
.css  .html  .sql  .graphql  .proto  .dockerfile

The following directory names are always skipped, even without --ignore:

node_modules  .git  .next  dist  __pycache__  .env  .env.local  .DS_Store

Any directory whose name starts with . is also skipped. .gitignore is not parsed — we don't read it. Use --ignore to add directory names of your own.

Examples

Import the current directory

memax import

Import a docs directory

memax import ./docs

Import with extra ignores

memax import ./repo --ignore "vendor,tmp,build"

Skip the "many files" confirmation

memax import ./large-dir --yes

`memax import status`

Prints the directories you've previously imported, when they last ran, and their last-run counts (scanned / pushed / skipped / errors).

memax import status

Behavior

Idempotent. Files are deduplicated server-side by content hash, so re-running memax import on an unchanged tree is a no-op.
Deletions are not propagated. Removing a file locally does not remove the corresponding memory from your hub — use memax delete for that.
Hub target. memax import does not pass a hub to the server — there's no --hub flag, .memax.yml hub: is not read by import, and memax hub switch does not route import writes. Imports land in whichever hub the server treats as the caller's default write target (normally the personal hub). To route specific content into a team hub, use memax push with --hub <slug> per file.
Watch mode is experimental. --watch is plumbed but the --help text flags it as "coming soon" while the UX is being validated.

On this page